BBN: Bilateral-Branch Network With Cumulative Learning for Long-Tailed Visual Recognition

TLDR

Long‑tailed visual recognition, where a few classes dominate the data while many others have few samples, is a challenging task that class‑rebalancing strategies such as re‑weighting and re‑sampling have been proposed to mitigate. This work proposes a Bilateral‑Branch Network (BBN) that simultaneously addresses representation learning and classifier learning by assigning each branch a distinct role. BBN incorporates a cumulative learning strategy that first captures universal patterns and then progressively focuses on tail classes. Extensive experiments on four benchmark datasets, including iNaturalist, show that BBN outperforms state‑of‑the‑art methods, won first place in the iNaturalist 2019 competition, and its effectiveness is validated by additional experiments, with the code released on GitHub.

Abstract

Our work focuses on tackling the challenging but natural visual recognition task of long-tailed data distribution (i.e., a few classes occupy most of the data, while most classes have rarely few samples). In the literature, class re-balancing strategies (e.g., re-weighting and re-sampling) are the prominent and effective methods proposed to alleviate the extreme imbalance for dealing with long-tailed problems. In this paper, we firstly discover that these re-balancing methods achieving satisfactory recognition accuracy owe to that they could significantly promote the classifier learning of deep networks. However, at the same time, they will unexpectedly damage the representative ability of the learned deep features to some extent. Therefore, we propose a unified Bilateral-Branch Network (BBN) to take care of both representation learning and classifier learning simultaneously, where each branch does perform its own duty separately. In particular, our BBN model is further equipped with a novel cumulative learning strategy, which is designed to first learn the universal patterns and then pay attention to the tail data gradually. Extensive experiments on four benchmark datasets, including the large-scale iNaturalist ones, justify that the proposed BBN can significantly outperform state-of-the-art methods. Furthermore, validation experiments can demonstrate both our preliminary discovery and effectiveness of tailored designs in BBN for long-tailed problems. Our method won the first place in the iNaturalist 2019 large scale species classification competition, and our code is open-source and available at https://github.com/Megvii-Nanjing/BBN.